
. 
. use data, clear

. 
. ***ANALYSIS*************************************************************
. 
. //Figure 1
. 
. xi: logit clicked i.condition voted dem rep q_age, r
i.condition       _Icondition_1-4     (naturally coded; _Icondition_1 omitted)

Iteration 0:   log pseudolikelihood = -11561.166  
Iteration 1:   log pseudolikelihood = -11039.184  
Iteration 2:   log pseudolikelihood = -10952.027  
Iteration 3:   log pseudolikelihood = -10951.823  
Iteration 4:   log pseudolikelihood = -10951.823  

Logistic regression                             Number of obs     =    140,144
                                                Wald chi2(7)      =    1173.45
                                                Prob > chi2       =     0.0000
Log pseudolikelihood = -10951.823               Pseudo R2         =     0.0527

-------------------------------------------------------------------------------
              |               Robust
      clicked |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------+----------------------------------------------------------------
_Icondition_2 |  -.5568326   .0584119    -9.53   0.000    -.6713178   -.4423474
_Icondition_3 |  -.5787835    .058791    -9.84   0.000    -.6940118   -.4635553
_Icondition_4 |   -.513082   .0576398    -8.90   0.000    -.6260539     -.40011
        voted |   1.015964   .0520986    19.50   0.000     .9138523    1.118075
          dem |  -.2574631   .0486146    -5.30   0.000     -.352746   -.1621802
          rep |   .2888707   .0791008     3.65   0.000     .1338359    .4439055
        q_age |   .4284563   .0219114    19.55   0.000     .3855107    .4714018
        _cons |  -5.466175   .0886013   -61.69   0.000     -5.63983    -5.29252
-------------------------------------------------------------------------------

. coefplot, drop(_I* _cons) xline(1) eform xtitle(" " "Click rate (odds ratio)") ///
>         mlabel format(%9.2f) mlabposition(12) mlabgap(*2) graphr(c(white))

. graph export "fig1.png", replace
(file fig1.png written in PNG format)

. 
. //Figure 2
. 
. cibar clicked, over1(condition) ///
>         barg(4) barc(gs1 gs7 gs7 gs7) ciopts(lc(gs10) lp(dash)) ///
>         gr(yti("Click rate (proportion)" " ") ysc(r(0 0.03)) ylab(0(0.01)0.03) ///
>         xlab(1 "Neither" 2 "Identity" 3 "Gratitude" 4 "Both") ///
>         graphr(c(white)) leg(off))

. graph export "fig2.png", replace
(file fig2.png written in PNG format)

. 
. logit clicked identity gratitude both voted, r

Iteration 0:   log pseudolikelihood = -11586.563  
Iteration 1:   log pseudolikelihood = -11284.754  
Iteration 2:   log pseudolikelihood = -11259.701  
Iteration 3:   log pseudolikelihood = -11259.658  
Iteration 4:   log pseudolikelihood = -11259.658  

Logistic regression                             Number of obs     =    140,189
                                                Wald chi2(4)      =     609.39
                                                Prob > chi2       =     0.0000
Log pseudolikelihood = -11259.658               Pseudo R2         =     0.0282

------------------------------------------------------------------------------
             |               Robust
     clicked |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
    identity |  -.5550225   .0581984    -9.54   0.000    -.6690893   -.4409558
   gratitude |  -.5771142   .0585687    -9.85   0.000    -.6919067   -.4623218
        both |  -.5086276   .0573519    -8.87   0.000    -.6210353   -.3962199
       voted |   1.057384   .0506711    20.87   0.000     .9580709    1.156698
       _cons |  -4.434841   .0535544   -82.81   0.000    -4.539806   -4.329877
------------------------------------------------------------------------------

. 
. 
. ***SUPPLEMENTAL INFORMATION*************************************************************
. 
. //Table S1: Summary statistics
. 
. summarize voted dem rep q_age male

    Variable |        Obs        Mean    Std. Dev.       Min        Max
-------------+---------------------------------------------------------
       voted |    140,189     .549822    .4975133          0          1
         dem |    140,189    .6377391    .4806554          0          1
         rep |    140,189    .0612316    .2397555          0          1
       q_age |    140,144    2.499836    1.118089          1          4
        male |     58,758    .5264134    .4993061          0          1

. 
. //Figure S1: Sensitivity analysis for Figure 1
. 
. eststo fig1: xi: logit clicked i.condition voted dem rep q_age, r
i.condition       _Icondition_1-4     (naturally coded; _Icondition_1 omitted)

Iteration 0:   log pseudolikelihood = -11561.166  
Iteration 1:   log pseudolikelihood = -11039.184  
Iteration 2:   log pseudolikelihood = -10952.027  
Iteration 3:   log pseudolikelihood = -10951.823  
Iteration 4:   log pseudolikelihood = -10951.823  

Logistic regression                             Number of obs     =    140,144
                                                Wald chi2(7)      =    1173.45
                                                Prob > chi2       =     0.0000
Log pseudolikelihood = -10951.823               Pseudo R2         =     0.0527

-------------------------------------------------------------------------------
              |               Robust
      clicked |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------+----------------------------------------------------------------
_Icondition_2 |  -.5568326   .0584119    -9.53   0.000    -.6713178   -.4423474
_Icondition_3 |  -.5787835    .058791    -9.84   0.000    -.6940118   -.4635553
_Icondition_4 |   -.513082   .0576398    -8.90   0.000    -.6260539     -.40011
        voted |   1.015964   .0520986    19.50   0.000     .9138523    1.118075
          dem |  -.2574631   .0486146    -5.30   0.000     -.352746   -.1621802
          rep |   .2888707   .0791008     3.65   0.000     .1338359    .4439055
        q_age |   .4284563   .0219114    19.55   0.000     .3855107    .4714018
        _cons |  -5.466175   .0886013   -61.69   0.000     -5.63983    -5.29252
-------------------------------------------------------------------------------

. eststo fig1_gender: xi: logit clicked i.condition voted dem rep q_age male, r
i.condition       _Icondition_1-4     (naturally coded; _Icondition_1 omitted)

Iteration 0:   log pseudolikelihood = -5907.2335  
Iteration 1:   log pseudolikelihood = -5624.2723  
Iteration 2:   log pseudolikelihood = -5583.5749  
Iteration 3:   log pseudolikelihood = -5583.3641  
Iteration 4:   log pseudolikelihood =  -5583.364  

Logistic regression                             Number of obs     =     58,758
                                                Wald chi2(8)      =     623.72
                                                Prob > chi2       =     0.0000
Log pseudolikelihood =  -5583.364               Pseudo R2         =     0.0548

-------------------------------------------------------------------------------
              |               Robust
      clicked |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
--------------+----------------------------------------------------------------
_Icondition_2 |  -.4760859   .0807949    -5.89   0.000    -.6344409   -.3177309
_Icondition_3 |  -.3781291   .0788214    -4.80   0.000    -.5326162   -.2236419
_Icondition_4 |  -.4580073   .0804335    -5.69   0.000     -.615654   -.3003605
        voted |   1.054126   .0759303    13.88   0.000     .9053056    1.202947
          dem |  -.2202148   .0665672    -3.31   0.001    -.3506841   -.0897455
          rep |   .2173901   .1061274     2.05   0.041     .0093842     .425396
        q_age |   .5685738   .0442909    12.84   0.000     .4817652    .6553824
         male |   .3962168   .0614979     6.44   0.000     .2756831    .5167505
        _cons |  -6.310321   .1725723   -36.57   0.000    -6.648557   -5.972086
-------------------------------------------------------------------------------

. coefplot (fig1,label(without gender) msiz(small)) ///
>                  (fig1_gender,label(with gender) m(triangle) msiz(small)), ///
>                  drop(_I* _cons) xline(1) eform xtitle(" " "Click rate (odds ratio)") ///
>                  mlabel format(%9.2f) mlabposition(12) mlabgap(*2) graphr(c(white))

. graph export "figS1.png", replace
(file figS1.png written in PNG format)

. eststo clear

. 
. //Figure S2: Balance test
. 
. forvalues c = 2/4 {
  2.         recode condition (1=`c')(`c'=1), gen(condition`c')
  3. }
(70045 differences between condition and condition2)
(70154 differences between condition and condition3)
(70194 differences between condition and condition4)

.         
. eststo balance12: mlogit condition voted dem rep q_age male, b(2) r rrr

Iteration 0:   log pseudolikelihood = -81454.485  
Iteration 1:   log pseudolikelihood = -81446.999  
Iteration 2:   log pseudolikelihood = -81446.999  

Multinomial logistic regression                 Number of obs     =     58,758
                                                Wald chi2(15)     =      15.07
                                                Prob > chi2       =     0.4467
Log pseudolikelihood = -81446.999               Pseudo R2         =     0.0001

---------------------------------------------------------------------------------
                |               Robust
      condition |        RRR   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
Neither         |
          voted |   .9501236   .0232025    -2.10   0.036     .9057188    .9967056
            dem |   .9920728   .0263084    -0.30   0.764     .9418264       1.045
            rep |    .966944   .0477664    -0.68   0.496      .877713    1.065246
          q_age |   1.024656    .013544     1.84   0.065     .9984517    1.051549
           male |   1.002178   .0236802     0.09   0.927     .9568243    1.049682
          _cons |   .9638142   .0450112    -0.79   0.430     .8795111    1.056198
----------------+----------------------------------------------------------------
Voter_identity  |  (base outcome)
----------------+----------------------------------------------------------------
Voter_gratitude |
          voted |   .9626663   .0235758    -1.55   0.120       .91755    1.010001
            dem |   1.019666   .0271628     0.73   0.465     .9677937    1.074318
            rep |   .9947014   .0491785    -0.11   0.914     .9028361    1.095914
          q_age |   1.011105   .0133371     0.84   0.402     .9852996    1.037586
           male |   .9877052   .0233317    -0.52   0.600     .9430183     1.03451
          _cons |   .9867369   .0457524    -0.29   0.773     .9010178    1.080611
----------------+----------------------------------------------------------------
Both            |
          voted |   .9455396   .0230635    -2.30   0.022     .9013994    .9918413
            dem |    1.01129   .0267607     0.42   0.671     .9601768    1.065124
            rep |   .9342022   .0464101    -1.37   0.171     .8475281     1.02974
          q_age |   1.009948   .0132596     0.75   0.451     .9842906    1.036273
           male |   1.021654   .0240528     0.91   0.363     .9755821    1.069901
          _cons |   1.006394   .0465075     0.14   0.890     .9192467    1.101802
---------------------------------------------------------------------------------

. eststo balance13: mlogit condition voted dem rep q_age male, b(3) r rrr

Iteration 0:   log pseudolikelihood = -81454.485  
Iteration 1:   log pseudolikelihood = -81446.999  
Iteration 2:   log pseudolikelihood = -81446.999  

Multinomial logistic regression                 Number of obs     =     58,758
                                                Wald chi2(15)     =      15.07
                                                Prob > chi2       =     0.4467
Log pseudolikelihood = -81446.999               Pseudo R2         =     0.0001

---------------------------------------------------------------------------------
                |               Robust
      condition |        RRR   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
Neither         |
          voted |    .986971   .0241313    -0.54   0.592     .9407899    1.035419
            dem |    .972939   .0258707    -1.03   0.302     .9235321    1.024989
            rep |   .9720947    .048128    -0.57   0.568     .8821978    1.071152
          q_age |   1.013403   .0134146     1.01   0.315     .9874489    1.040039
           male |   1.014653   .0239428     0.62   0.538     .9687949    1.062683
          _cons |   .9767692   .0455238    -0.50   0.614     .8914981    1.070197
----------------+----------------------------------------------------------------
Voter_identity  |
          voted |   1.038782   .0254399     1.55   0.120     .9900981    1.089859
            dem |   .9807134   .0261252    -0.73   0.465     .9308228    1.033278
            rep |   1.005327   .0497038     0.11   0.914     .9124803    1.107621
          q_age |   .9890171   .0130458    -0.84   0.402     .9637756     1.01492
           male |   1.012448   .0239162     0.52   0.600     .9666416    1.060425
          _cons |   1.013441   .0469907     0.29   0.773     .9254024    1.109856
----------------+----------------------------------------------------------------
Voter_gratitude |  (base outcome)
----------------+----------------------------------------------------------------
Both            |
          voted |   .9822092   .0239875    -0.74   0.462      .936302    1.030367
            dem |   .9917855   .0263147    -0.31   0.756     .9415278    1.044726
            rep |   .9391785   .0467649    -1.26   0.208     .8518516    1.035458
          q_age |   .9988554   .0131317    -0.09   0.931     .9734464    1.024928
           male |   1.034371    .024319     1.44   0.151     .9877883    1.083151
          _cons |   1.019921   .0470311     0.43   0.669     .9317844    1.116394
---------------------------------------------------------------------------------

. eststo balance14: mlogit condition voted dem rep q_age male, b(4) r rrr

Iteration 0:   log pseudolikelihood = -81454.485  
Iteration 1:   log pseudolikelihood = -81446.999  
Iteration 2:   log pseudolikelihood = -81446.999  

Multinomial logistic regression                 Number of obs     =     58,758
                                                Wald chi2(15)     =      15.07
                                                Prob > chi2       =     0.4467
Log pseudolikelihood = -81446.999               Pseudo R2         =     0.0001

---------------------------------------------------------------------------------
                |               Robust
      condition |        RRR   Std. Err.      z    P>|z|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
Neither         |
          voted |   1.004848   .0244698     0.20   0.843     .9580147    1.053971
            dem |   .9809975   .0259099    -0.73   0.468     .9315069    1.033117
            rep |   1.035048   .0514952     0.69   0.489     .9388839    1.141061
          q_age |   1.014564   .0133665     1.10   0.272     .9887016    1.041103
           male |   .9809375   .0230684    -0.82   0.413     .9367505    1.027209
          _cons |   .9576911   .0444869    -0.93   0.352     .8743499    1.048976
----------------+----------------------------------------------------------------
Voter_identity  |
          voted |   1.057597   .0257968     2.30   0.022     1.008226    1.109386
            dem |   .9888362   .0261665    -0.42   0.671      .938858    1.041475
            rep |   1.070432   .0531779     1.37   0.171     .9711189    1.179902
          q_age |   .9901504   .0129997    -0.75   0.451     .9649965     1.01596
           male |   .9788053    .023044    -0.91   0.363      .934666    1.025029
          _cons |    .993647   .0459185    -0.14   0.890     .9076038    1.087847
----------------+----------------------------------------------------------------
Voter_gratitude |
          voted |   1.018113   .0248643     0.74   0.462     .9705278    1.068031
            dem |   1.008283   .0267524     0.31   0.756     .9571889    1.062104
            rep |    1.06476    .053018     1.26   0.208     .9657565    1.173913
          q_age |   1.001146   .0131618     0.09   0.931     .9756787    1.027278
           male |    .966771   .0227297    -1.44   0.151     .9232326    1.012363
          _cons |   .9804682   .0452119    -0.43   0.669      .895741     1.07321
----------------+----------------------------------------------------------------
Both            |  (base outcome)
---------------------------------------------------------------------------------

. eststo balance23: mlogit condition2 voted dem rep q_age male, b(3) r rrr

Iteration 0:   log pseudolikelihood = -81454.485  
Iteration 1:   log pseudolikelihood = -81446.999  
Iteration 2:   log pseudolikelihood = -81446.999  

Multinomial logistic regression                 Number of obs     =     58,758
                                                Wald chi2(15)     =      15.07
                                                Prob > chi2       =     0.4467
Log pseudolikelihood = -81446.999               Pseudo R2         =     0.0001

------------------------------------------------------------------------------
             |               Robust
  condition2 |        RRR   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
1            |
       voted |   1.038782   .0254399     1.55   0.120     .9900981    1.089859
         dem |   .9807134   .0261252    -0.73   0.465     .9308228    1.033278
         rep |   1.005327   .0497038     0.11   0.914     .9124803    1.107621
       q_age |   .9890171   .0130458    -0.84   0.402     .9637756     1.01492
        male |   1.012448   .0239162     0.52   0.600     .9666416    1.060425
       _cons |   1.013441   .0469907     0.29   0.773     .9254024    1.109856
-------------+----------------------------------------------------------------
2            |
       voted |    .986971   .0241313    -0.54   0.592     .9407899    1.035419
         dem |    .972939   .0258707    -1.03   0.302     .9235321    1.024989
         rep |   .9720947    .048128    -0.57   0.568     .8821978    1.071152
       q_age |   1.013403   .0134146     1.01   0.315     .9874489    1.040039
        male |   1.014653   .0239428     0.62   0.538     .9687949    1.062683
       _cons |   .9767692   .0455238    -0.50   0.614     .8914981    1.070197
-------------+----------------------------------------------------------------
3            |  (base outcome)
-------------+----------------------------------------------------------------
4            |
       voted |   .9822092   .0239875    -0.74   0.462      .936302    1.030367
         dem |   .9917855   .0263147    -0.31   0.756     .9415278    1.044726
         rep |   .9391785   .0467649    -1.26   0.208     .8518516    1.035458
       q_age |   .9988554   .0131317    -0.09   0.931     .9734464    1.024928
        male |   1.034371    .024319     1.44   0.151     .9877883    1.083151
       _cons |   1.019921   .0470311     0.43   0.669     .9317844    1.116394
------------------------------------------------------------------------------

. eststo balance24: mlogit condition2 voted dem rep q_age male, b(4) r rrr

Iteration 0:   log pseudolikelihood = -81454.485  
Iteration 1:   log pseudolikelihood = -81446.999  
Iteration 2:   log pseudolikelihood = -81446.999  

Multinomial logistic regression                 Number of obs     =     58,758
                                                Wald chi2(15)     =      15.07
                                                Prob > chi2       =     0.4467
Log pseudolikelihood = -81446.999               Pseudo R2         =     0.0001

------------------------------------------------------------------------------
             |               Robust
  condition2 |        RRR   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
1            |
       voted |   1.057597   .0257968     2.30   0.022     1.008226    1.109386
         dem |   .9888362   .0261665    -0.42   0.671      .938858    1.041475
         rep |   1.070432   .0531779     1.37   0.171     .9711189    1.179902
       q_age |   .9901504   .0129997    -0.75   0.451     .9649965     1.01596
        male |   .9788053    .023044    -0.91   0.363      .934666    1.025029
       _cons |    .993647   .0459185    -0.14   0.890     .9076038    1.087847
-------------+----------------------------------------------------------------
2            |
       voted |   1.004848   .0244698     0.20   0.843     .9580147    1.053971
         dem |   .9809975   .0259099    -0.73   0.468     .9315069    1.033117
         rep |   1.035048   .0514952     0.69   0.489     .9388839    1.141061
       q_age |   1.014564   .0133665     1.10   0.272     .9887016    1.041103
        male |   .9809375   .0230684    -0.82   0.413     .9367505    1.027209
       _cons |   .9576911   .0444869    -0.93   0.352     .8743499    1.048976
-------------+----------------------------------------------------------------
3            |
       voted |   1.018113   .0248643     0.74   0.462     .9705278    1.068031
         dem |   1.008283   .0267524     0.31   0.756     .9571889    1.062104
         rep |    1.06476    .053018     1.26   0.208     .9657565    1.173913
       q_age |   1.001146   .0131618     0.09   0.931     .9756787    1.027278
        male |    .966771   .0227297    -1.44   0.151     .9232326    1.012363
       _cons |   .9804682   .0452119    -0.43   0.669      .895741     1.07321
-------------+----------------------------------------------------------------
4            |  (base outcome)
------------------------------------------------------------------------------

. eststo balance34: mlogit condition3 voted dem rep q_age male, b(4) r rrr

Iteration 0:   log pseudolikelihood = -81454.485  
Iteration 1:   log pseudolikelihood = -81446.999  
Iteration 2:   log pseudolikelihood = -81446.999  

Multinomial logistic regression                 Number of obs     =     58,758
                                                Wald chi2(15)     =      15.07
                                                Prob > chi2       =     0.4467
Log pseudolikelihood = -81446.999               Pseudo R2         =     0.0001

------------------------------------------------------------------------------
             |               Robust
  condition3 |        RRR   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
1            |
       voted |   1.018113   .0248643     0.74   0.462     .9705278    1.068031
         dem |   1.008283   .0267524     0.31   0.756     .9571889    1.062104
         rep |    1.06476    .053018     1.26   0.208     .9657565    1.173913
       q_age |   1.001146   .0131618     0.09   0.931     .9756787    1.027278
        male |    .966771   .0227297    -1.44   0.151     .9232326    1.012363
       _cons |   .9804682   .0452119    -0.43   0.669      .895741     1.07321
-------------+----------------------------------------------------------------
2            |
       voted |   1.057597   .0257968     2.30   0.022     1.008226    1.109386
         dem |   .9888362   .0261665    -0.42   0.671      .938858    1.041475
         rep |   1.070432   .0531779     1.37   0.171     .9711189    1.179902
       q_age |   .9901504   .0129997    -0.75   0.451     .9649965     1.01596
        male |   .9788053    .023044    -0.91   0.363      .934666    1.025029
       _cons |    .993647   .0459185    -0.14   0.890     .9076038    1.087847
-------------+----------------------------------------------------------------
3            |
       voted |   1.004848   .0244698     0.20   0.843     .9580147    1.053971
         dem |   .9809975   .0259099    -0.73   0.468     .9315069    1.033117
         rep |   1.035048   .0514952     0.69   0.489     .9388839    1.141061
       q_age |   1.014564   .0133665     1.10   0.272     .9887016    1.041103
        male |   .9809375   .0230684    -0.82   0.413     .9367505    1.027209
       _cons |   .9576911   .0444869    -0.93   0.352     .8743499    1.048976
-------------+----------------------------------------------------------------
4            |  (base outcome)
------------------------------------------------------------------------------

. 
. coefplot(balance12,label(Neither vs. Identity) ciop(lc(gs0)) mc(gs0)) ///
>                 (balance13,label(Neither vs. Gratitude) ciop(lc(gs2)) mc(gs2)) ///
>                 (balance14,label(Neither vs. Both) ciop(lc(gs4)) mc(gs4)) ///
>                 (balance23,label(Identity vs. Gratitude) ciop(lc(gs6)) mc(gs6)) ///
>                 (balance24,label(Identity vs. Both) ciop(lc(gs8)) mc(gs8)) ///
>                 (balance34,label(Gratitude vs. Both) ciop(lc(gs10)) mc(gs10)), ///
>                 drop(_cons) xline(1) eform xtitle(Odds ratio) legend(cols(1) ring(0) pos(5) //
> /
>                 bm(none) m(zero) si(small)) graphr(c(white))
(note:  named style none not found in class margin, default attributes used)

. graph export "figS2.png", replace
(file figS2.png written in PNG format)

. eststo clear

. 
. //Figure S3: Sensitivity analysis for covariate adjustment
. 
. eststo noadjust: logit clicked identity gratitude both, r // without adjustment

Iteration 0:   log pseudolikelihood = -11586.563  
Iteration 1:   log pseudolikelihood =  -11517.35  
Iteration 2:   log pseudolikelihood = -11514.732  
Iteration 3:   log pseudolikelihood = -11514.729  
Iteration 4:   log pseudolikelihood = -11514.729  

Logistic regression                             Number of obs     =    140,189
                                                Wald chi2(3)      =     151.82
                                                Prob > chi2       =     0.0000
Log pseudolikelihood = -11514.729               Pseudo R2         =     0.0062

------------------------------------------------------------------------------
             |               Robust
     clicked |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
    identity |  -.5505163   .0581126    -9.47   0.000     -.664415   -.4366177
   gratitude |  -.5730226   .0584591    -9.80   0.000    -.6876003    -.458445
        both |  -.5049462   .0572375    -8.82   0.000    -.6171296   -.3927628
       _cons |  -3.731823   .0353158  -105.67   0.000    -3.801041   -3.662606
------------------------------------------------------------------------------

. eststo adjust: logit clicked identity gratitude both voted, r // with adjustment

Iteration 0:   log pseudolikelihood = -11586.563  
Iteration 1:   log pseudolikelihood = -11284.754  
Iteration 2:   log pseudolikelihood = -11259.701  
Iteration 3:   log pseudolikelihood = -11259.658  
Iteration 4:   log pseudolikelihood = -11259.658  

Logistic regression                             Number of obs     =    140,189
                                                Wald chi2(4)      =     609.39
                                                Prob > chi2       =     0.0000
Log pseudolikelihood = -11259.658               Pseudo R2         =     0.0282

------------------------------------------------------------------------------
             |               Robust
     clicked |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
    identity |  -.5550225   .0581984    -9.54   0.000    -.6690893   -.4409558
   gratitude |  -.5771142   .0585687    -9.85   0.000    -.6919067   -.4623218
        both |  -.5086276   .0573519    -8.87   0.000    -.6210353   -.3962199
       voted |   1.057384   .0506711    20.87   0.000     .9580709    1.156698
       _cons |  -4.434841   .0535544   -82.81   0.000    -4.539806   -4.329877
------------------------------------------------------------------------------

. 
. coefplot (noadjust,label(No adjustment)) (adjust,label(Adjustment)), ///
>         drop(voted _cons) xline(1) eform xtitle(Odds ratio) legend(cols(1) ring(0) pos(3) ///
>         bm(none) m(zero) si(small)) graphr(c(white))
(note:  named style none not found in class margin, default attributes used)

. graph export "figS3.png", replace
(file figS3.png written in PNG format)

. eststo clear

. 
. //Table S2: Heterogeneous effects by primary voter
. 
. eststo: logit clicked identity##i.voted gratitude both dem rep q_age male, r

Iteration 0:   log pseudolikelihood = -5907.2335  
Iteration 1:   log pseudolikelihood =  -5623.639  
Iteration 2:   log pseudolikelihood = -5580.8493  
Iteration 3:   log pseudolikelihood = -5580.4255  
Iteration 4:   log pseudolikelihood = -5580.4245  
Iteration 5:   log pseudolikelihood = -5580.4245  

Logistic regression                             Number of obs     =     58,758
                                                Wald chi2(9)      =     608.17
                                                Prob > chi2       =     0.0000
Log pseudolikelihood = -5580.4245               Pseudo R2         =     0.0553

------------------------------------------------------------------------------------------
                         |               Robust
                 clicked |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------------------+----------------------------------------------------------------
              1.identity |  -.8758697   .1934008    -4.53   0.000    -1.254928   -.4968111
                         |
                   voted |
  Voted in 2016 primary  |   .9670709   .0828051    11.68   0.000     .8047759    1.129366
                         |
          identity#voted |
1#Voted in 2016 primary  |   .4773092   .2042303     2.34   0.019     .0770251    .8775933
                         |
               gratitude |  -.3779283   .0787637    -4.80   0.000    -.5323023   -.2235543
                    both |  -.4579967   .0803759    -5.70   0.000    -.6155306   -.3004629
                     dem |  -.2212825   .0665476    -3.33   0.001    -.3517134   -.0908517
                     rep |   .2166459   .1061172     2.04   0.041     .0086601    .4246317
                   q_age |   .5681039   .0442874    12.83   0.000     .4813023    .6549056
                    male |   .3954044   .0614954     6.43   0.000     .2748756    .5159332
                   _cons |  -6.237593    .174541   -35.74   0.000    -6.579687   -5.895499
------------------------------------------------------------------------------------------
(est1 stored)

. estadd local controls "Yes"

added macro:
           e(controls) : "Yes"

. estadd local cfx "Yes"

added macro:
                e(cfx) : "Yes"

.         margins, dydx(identity) over(voted)

Average marginal effects                        Number of obs     =     58,758
Model VCE    : Robust

Expression   : Pr(clicked), predict()
dy/dx w.r.t. : 1.identity
over         : voted

----------------------------------------------------------------------------------------
                       |            Delta-method
                       |      dy/dx   Std. Err.      z    P>|z|     [95% Conf. Interval]
-----------------------+----------------------------------------------------------------
1.identity             |
                 voted |
         Did not vote  |  -.0065423    .001186    -5.52   0.000    -.0088668   -.0042179
Voted in 2016 primary  |  -.0100249   .0019988    -5.02   0.000    -.0139426   -.0061073
----------------------------------------------------------------------------------------
Note: dy/dx for factor levels is the discrete change from the base level.

. 
. eststo: logit clicked identity gratitude##i.voted both dem rep q_age male, r

Iteration 0:   log pseudolikelihood = -5907.2335  
Iteration 1:   log pseudolikelihood = -5624.3657  
Iteration 2:   log pseudolikelihood = -5583.4044  
Iteration 3:   log pseudolikelihood = -5583.1879  
Iteration 4:   log pseudolikelihood = -5583.1879  

Logistic regression                             Number of obs     =     58,758
                                                Wald chi2(9)      =     622.23
                                                Prob > chi2       =     0.0000
Log pseudolikelihood = -5583.1879               Pseudo R2         =     0.0549

------------------------------------------------------------------------------------------
                         |               Robust
                 clicked |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------------------+----------------------------------------------------------------
                identity |  -.4758183    .080765    -5.89   0.000    -.6341147   -.3175218
             1.gratitude |  -.4644164   .1674311    -2.77   0.006    -.7925754   -.1362575
                         |
                   voted |
  Voted in 2016 primary  |   1.029923   .0855512    12.04   0.000     .8622459      1.1976
                         |
         gratitude#voted |
1#Voted in 2016 primary  |   .1059146   .1792839     0.59   0.555    -.2454753    .4573045
                         |
                    both |  -.4580009   .0804177    -5.70   0.000    -.6156167   -.3003852
                     dem |  -.2200089   .0665667    -3.31   0.001    -.3504771   -.0895407
                     rep |   .2177431   .1061412     2.05   0.040     .0097101    .4257761
                   q_age |    .568653   .0442834    12.84   0.000     .4818592    .6554468
                    male |   .3963143   .0614969     6.44   0.000     .2757826     .516846
                   _cons |  -6.291297   .1766887   -35.61   0.000      -6.6376   -5.944993
------------------------------------------------------------------------------------------
(est2 stored)

. estadd local controls "Yes"

added macro:
           e(controls) : "Yes"

. estadd local cfx "Yes"

added macro:
                e(cfx) : "Yes"

.         margins, dydx(gratitude) over(voted)

Average marginal effects                        Number of obs     =     58,758
Model VCE    : Robust

Expression   : Pr(clicked), predict()
dy/dx w.r.t. : 1.gratitude
over         : voted

----------------------------------------------------------------------------------------
                       |            Delta-method
                       |      dy/dx   Std. Err.      z    P>|z|     [95% Conf. Interval]
-----------------------+----------------------------------------------------------------
1.gratitude            |
                 voted |
         Did not vote  |  -.0038926   .0012796    -3.04   0.002    -.0064006   -.0013845
Voted in 2016 primary  |  -.0091003    .002009    -4.53   0.000     -.013038   -.0051627
----------------------------------------------------------------------------------------
Note: dy/dx for factor levels is the discrete change from the base level.

. 
. eststo: logit clicked identity gratitude both##i.voted dem rep q_age male, r

Iteration 0:   log pseudolikelihood = -5907.2335  
Iteration 1:   log pseudolikelihood = -5625.0548  
Iteration 2:   log pseudolikelihood = -5583.2688  
Iteration 3:   log pseudolikelihood =   -5583.06  
Iteration 4:   log pseudolikelihood = -5583.0599  

Logistic regression                             Number of obs     =     58,758
                                                Wald chi2(9)      =     627.80
                                                Prob > chi2       =     0.0000
Log pseudolikelihood = -5583.0599               Pseudo R2         =     0.0549

------------------------------------------------------------------------------------------
                         |               Robust
                 clicked |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------------------+----------------------------------------------------------------
                identity |   -.476433    .080797    -5.90   0.000    -.6347923   -.3180737
               gratitude |  -.3782068   .0788399    -4.80   0.000    -.5327302   -.2236834
                  1.both |   -.347064   .1626775    -2.13   0.033    -.6659061   -.0282219
                         |
                   voted |
  Voted in 2016 primary  |   1.085783   .0865909    12.54   0.000     .9160675    1.255498
                         |
              both#voted |
1#Voted in 2016 primary  |  -.1381528   .1757771    -0.79   0.432    -.4826696     .206364
                         |
                     dem |  -.2205035   .0665696    -3.31   0.001    -.3509776   -.0900294
                     rep |   .2172953   .1061271     2.05   0.041       .00929    .4253006
                   q_age |   .5684203   .0442944    12.83   0.000     .4816048    .6552357
                    male |   .3961779   .0615011     6.44   0.000      .275638    .5167178
                   _cons |  -6.335241   .1761594   -35.96   0.000    -6.680507   -5.989975
------------------------------------------------------------------------------------------
(est3 stored)

. estadd local controls "Yes"

added macro:
           e(controls) : "Yes"

. estadd local cfx "Yes"

added macro:
                e(cfx) : "Yes"

.         margins, dydx(both) over(voted)

Average marginal effects                        Number of obs     =     58,758
Model VCE    : Robust

Expression   : Pr(clicked), predict()
dy/dx w.r.t. : 1.both
over         : voted

----------------------------------------------------------------------------------------
                       |            Delta-method
                       |      dy/dx   Std. Err.      z    P>|z|     [95% Conf. Interval]
-----------------------+----------------------------------------------------------------
1.both                 |
                 voted |
         Did not vote  |  -.0029919   .0013119    -2.28   0.023    -.0055632   -.0004206
Voted in 2016 primary  |  -.0119587   .0019552    -6.12   0.000    -.0157909   -.0081265
----------------------------------------------------------------------------------------
Note: dy/dx for factor levels is the discrete change from the base level.

. 
. esttab using "tabS2.rtf", replace compress onecell label se pr2 ///
>         drop(identity gratitude both male dem rep q_age _cons) nobaselevels ///
>         scalar("controls Controls" "cfx Condition-FE") ///
>         rename(1.gratitude#1.voted "Interaction", 1.identity#1.voted "Interaction", 1.both#1.v
> oted "Interaction", ///
>                   1.identity "Treatment", 1.gratitude "Treatment", 1.both "Treatment") ///
>         mtitles ("Identity" "Gratitude" "Both") 
(note: file tabS2.rtf not found)
(output written to tabS2.rtf)

. eststo clear

. 
. cap log close
